The 18% Water You Forget About

The 18% Water You Forget About

Butter is roughly 80% fat, 18% water, and 2% milk solids — and almost every successful or failed butter substitute comes down to what happens to that 18%. The best swaps either match it (ghee with 25% less to compensate for its zero water, lard at 0.875:1 for pastry) or replace its function (cream at half-volume for richness, olive oil at 0.75:1 for baking). The substitutes that fail — coconut oil in pie crust, heavy cream in frosting — fail because the water either disappears or refuses to leave at the right time.

The water you don't see, and what it actually does

Pull a stick of butter from the fridge and you're holding a stable emulsion: about 80% milkfat with 18% water suspended as microscopic droplets, bound by 2% milk solids acting as surfactant. That number — 18% — is the hidden lever behind almost every recipe that calls for butter by name. When a cake batter creams light and tall, when a pie crust shatters into leaves, when a beurre blanc holds together long enough to plate, the water is doing structural work that the fat alone cannot do. European-style butters push toward 82-84% fat (closer to 16% water), American supermarket butters typically sit at 80% fat (closer to 18% water), and so-called spreadable butters can drop to 60% fat with the rest made up of water and added oils. None of these distinctions are cosmetic; they shift how a recipe behaves by enough that pastry chefs deliberately specify the brand.

In a creamed cake, you beat butter with sugar until the sugar crystals shear pockets into the fat and trap air. The water is barely a player yet. But once eggs and flour go in, the batter heats, and the water inside each butter pocket flashes to steam — that steam is what pushes against the gluten network and lifts the crumb alongside the chemical leavener. Take the water away (use ghee, which is essentially 100% fat) and the crumb gets denser and slightly greasy; recipes calling for ghee in cakes typically increase liquid elsewhere or reduce the fat by 25% to compensate, which matches the database guidance to use 25% less ghee — no water content when swapping into baked goods. The interaction with eggs is part of what makes the substitution non-trivial — egg whites can carry some of the missing aeration, but only if the recipe is rebalanced toward whipped whites rather than whole eggs folded in.

In a laminated dough — pie crust, croissant, rough puff — the water does even more dramatic work. Cold butter sheets between flour layers, then melts in the oven. The fat coats the gluten and fries the starch; the water boils through it and inflates each layer like a tiny balloon. This is why lard at 0.875:1 produces flaky pastry crust with the explicit caveat that lard "has no water content unlike butter's 15-20%" — the swap works specifically because lard's plastic, malleable solid fat behaves like butter at lamination temperature, but the cook has to add steam from elsewhere (often a wetter dough, sometimes a vodka-water binder) to recover the lift the missing water would have provided. The window where butter is plastic — about 60-68°F — is narrow enough that professional bakers refrigerate their lamination station; the water has to stay inside the butter as droplets, not migrate into the dough, until it's ready to flash.

In a sauce, the water is the emulsion's hostage. A hot pan with butter melts the fat first, but the water rolls into solution with whatever's also liquid — wine, stock, lemon. The milk solids stabilize that emulsion until temperature climbs past about 190°F, at which point the proteins denature and the sauce breaks. Cooking applicability data ranks butter at an average score of 3.93 for cooking and 3.63 for sauce — high, but not the savory-application 4.19 — and the gap is explained almost entirely by sauce cooks who burned the water off too fast and watched their pan sauce split. A monté-au-beurre finish to a pan reduction works because cold butter's water phase is what carries the casein and lecithin into the sauce as a stable emulsion; melt the butter first and you've already broken the emulsion in the saucepan you used to melt it.

That same 18% explains the next section: every "butter alternative" has to be classified by where its water is, how much it has, and whether it can release that water at the same temperature butter does. Once that's the lens, the swap matrix stops being a list and becomes a chart of moisture profiles.

The substitutes that share the water — and the ones that fight it

Group one: substitutes whose water content is roughly equivalent to butter's, or which release moisture in the same window. The cleanest example is salted butter: identical 80/18/2 split, same melt curve, with the only adjustment being a pinch of salt per stick removed from the recipe. The database scores it 100/100 function-match at a 1:1 ratio precisely because there's nothing chemical to manage. Stick butter, whipped butter, and cultured butter all live in this group; whipped butter is the only one with a real ratio twist (3:2 in the database — use three tablespoons whipped to replace two tablespoons regular — because the air folded into whipped butter has displaced about a third of the volume but none of the fat or water). Cultured butter, which has been fermented before churning, behaves identically in baking but adds a lactic-acid tang that can either complement a recipe (shortbread, croissants) or fight it (a delicate génoise where you wanted neutrality).

Group two: substitutes with no water that have to be metered down or supplemented. Ghee is the cleanest case. Pure butterfat, smoke point near 485°F, dairy-free of casein for lactose-sensitive cooks. The database lists ghee at a 1:1 ratio with a 100/100 function-match, but the warnings layer adds two crucial details: ghee browns faster in the pan because its higher smoke point means no water is buffering the temperature, and the clarified butter (ghee) butter entry says to use 25% less when the application is moisture-dependent. In practice this means: 1:1 for sautéing, basting, finishing — anywhere the water in regular butter would have boiled off in seconds anyway.

But for cakes, biscuits, pie crusts, and anything else where the steam matters, drop the ghee by a quarter and add liquid back from the recipe's other components. Lard at 1:1 (function-match 100/100) and at 0.875:1 for pastry (function-match 80/100) follows the same logic — it has no water, so when you want flake, you back the fat off slightly to leave room for a wetter dough. Beef tallow, schmaltz, and bacon fat all behave the same way mechanically — pure fat, with the further constraint of carrying flavor that drives them to savory applications only.

Group three: substitutes with their own water but at different concentrations. Cream — heavy, light, half-and-half — sits here. Heavy cream is roughly 36-40% fat and 60% water; half-and-half is about 12% fat and 87% water. The database scores cream at 100/100 with a striking ratio: 1:1.5 cup, with the explicit note "use half the amount as butter, adds silky mouthfeel," and half and half at 1:0.875 cup with the warning that it's much thinner and works in sauces and soups but not where solid fat is needed. The reason isn't the fat percentage — it's that liquid cream cannot hold its own structure. The texture warning liquid cream won't create flaky layers is exact: if the recipe needs the substitute to behave like a solid until it hits the oven, cream and half-and-half are out. They get redirected to applications where butter's role was richness and emulsion, not architecture.

The deepest cousin of butter in this group is heavy cream whipped to soft peaks for frosting (a 1:0.333 cup ratio at 80/100 function-match), where you're using the cream's own emulsified water to mimic creamed butter's air pockets — but the database flags it as richer than butter but adds no structure, meaning a buttercream substituted with whipped cream will frost beautifully and collapse in two hours on a warm counter. For applications where you specifically need cream's water-handling profile — Alfredo, ganache, panna cotta — the full breakdown of heavy cream swaps walks through which thickeners can stand in. Sour cream and Greek yogurt also belong to this group, with the further wrinkle that their lactic acid weakens gluten — useful in muffins and quick breads, deadly in laminated doughs.

Group four: liquid oils, which have no water at all. Olive oil at 0.75:1 cup is the canonical example: 100/100 function-match, with explicit fitness for baking and finishing. But the structural warnings stack up fast. Liquid oil cannot create flaky pie layers appears for both olive oil and butter oil; liquid oil will not hold frosting shape appears twice. The flavor warning fruity olive flavor changes baked goods is its own constraint, which is why a mild refined oil works better in a chiffon and an extra-virgin shines in a focaccia or savory bread.

Vegetable oil follows the same chemistry without the flavor; the vegetable oil journal piece covers when to use it instead of melted butter and when the missing milk solids matter. The unifying principle: oils replace butter only in applications where the butter was already going to melt — quick breads, muffins, brownies, sautés finished with a knob — and only at 75-80% of the volume because pure fat is more efficient than fat-plus-water by mass. The 25% reduction has a second purpose: it leaves room for an extra tablespoon or two of milk or buttermilk to recover the moisture and the dairy proteins, which is why most published "oil-substitute" cake recipes also bump the liquid by a similar percentage.

The transition from group four to the failure modes is direct. Once an oil is liquid at room temperature, it cannot store water or air. Once that's true, you can predict every recipe it will ruin.

Where the swap fails: a chart of broken water

Coconut oil sits in a strange position. Solid below 76°F, liquid above. The texture warning solid below 76F — measure melted for baking is operationally important — if your kitchen is cool, the coconut oil is acting like a fat with no water; if your kitchen is warm, it's acting like a flavorless oil with no water. Either way it has zero of butter's 18%, and the structural warnings carry: liquid oil cannot create flaky pie layers. The only application where coconut oil and butter are clean substitutes is in melted-butter recipes — pound cake variants, blondies, the all-in-one wet-method muffins — where the fat's job is moisture-binding and richness, not aeration.

In a pie crust, coconut oil produces a crumbly, sandy texture rather than a flaky one, because there's no steam to inflate the layers and the solid coconut oil melts at a lower, narrower window than butter, leaving the layers pre-merged before the gluten can set. The dish-page data ranks pie crust as one of the top 14 substitutions tracked, and the reason the table has so many entries is that almost every common butter substitute fails it in a different way. Refined coconut oil removes the coconut flavor but doesn't recover any of the missing water; unrefined coconut oil keeps the flavor and gets flagged with unrefined coconut oil adds coconut taste, which is fine for a coconut-forward cookie and ruinous for a buttery French shortbread.

Frosting failures are the second predictable cluster. American buttercream relies on butter's solid-at-room-temperature structure to hold piped shape; the milk solids and water emulsify the powdered sugar into a stable foam. Heavy cream whipped, vegetable shortening, and softened cream cheese all produce frostings that taste right and pipe wrong. The cream cheese case is interesting: it brings water of its own (about 55%) plus protein and acid, which is why a cream-cheese frosting at the standard 1:1 with butter actually works — but a 100% cream cheese substitution sags within an hour because the water-to-fat ratio has flipped. The cream cheese piece breaks down the dilution math. Whipped frostings rebuilt around a stabilized whipped cream (with gelatin or mascarpone added) recover some shape but never the same pipe-friendliness, because the water in cream is held by far weaker bonds than the water trapped inside butter's microcrystalline lattice.

Margarine and shortening occupy a special middle ground. Margarine is engineered to mimic butter's water content (around 16-18%), and the texture warning softer cookies — less crisp edges is the only consistent failure note — softer because most margarines have slightly more water and softer plastic fat, so cookies spread further before setting. Shortening is closer to lard in fat profile but with zero water and the warning no rich dairy flavor — just neutral fat. Use shortening in pie crust and you get the flake (because the fat is plastic at room temperature and laminates well) without the browning or flavor. The classic American pie-crust trick — half butter, half shortening — exists precisely to split the difference: butter's water makes the lift and browning, shortening's plasticity makes the structure forgiving. A 50/50 blend predictably outperforms either alone for cooks who don't yet have the touch to handle pure butter without overworking it.

Then there are the savory failures. Chicken fat, duck fat, and bacon grease all carry warning notes about flavor — savory poultry flavor — not for sweet baking, strong duck flavor — savory dishes only. Mechanically these are excellent butter substitutes for sautés, roasted vegetables, and rich savory pastries (a duck-fat biscuit is a real and good thing), but the savory-applications-only constraint is absolute. The database's savory applicability score of 4.19 for butter sets the ceiling, and animal fats hit it; the dessert score of 2.85 sets the floor, and animal fats fall through it. Butter's drink-applicability score of 1.63 — the lowest of all use cases — is a quiet reminder that no fat substitution rule survives contact with a cocktail; the only "buttered drinks" the database scores are hot toddies and bulletproof coffee, and there butter is itself the substitute for a cream or oil that was failing.

The lesson across these failures is consistent: when a butter substitute breaks a recipe, it's almost always because the water moved at the wrong temperature, in the wrong volume, or not at all. Once you can predict where the water is, you can predict where the swap holds.

Adjusting the recipe around the swap

Every swap that diverges from butter's water profile demands a recipe-level correction, not just a 1:1 substitution. There are four levers, and they're all about restoring or accommodating moisture.

The first lever is supplemental water. The database guidance for butter oil is precise: use 20% less butter oil and add water — a five-tablespoon swap becomes four tablespoons of butter oil plus one tablespoon of water, recombining the 80/18 ratio in the bowl rather than relying on a pre-emulsified product. This works for ghee in pie crust (use the 25%-less rule plus an ice-water bump), for shortening in biscuits (cut in the shortening, then add an extra tablespoon of buttermilk), and for any 100%-fat substitute in a recipe where the lift depends on steam. The water has to be cold and added at the end, the same way it would be in a butter-based dough; warm water hydrates the gluten too aggressively and trades flake for chew.

The second lever is sugar adjustment. Butter's water dissolves a portion of the recipe's sugar; once dissolved, that sugar can't recrystallize into the same sandy creaming structure. Swap to a no-water fat and the sugar stays granular longer — which is why ghee cakes and shortening cookies often benefit from finer-grained sugar (caster, or pulse-processed granulated). The role of granulated sugar in moisture management is its own subject — see the granulated sugar journal piece for the full breakdown of sugar's hygroscopic behavior — but as a butter-swap heuristic: if you remove water by changing fat, increase the sugar's surface area to hold what moisture remains. Switching part of the sugar to brown sugar, which carries about 2-3% molasses moisture of its own, is another way to put water back into a recipe that lost it through the fat.

The third lever is leavening. Butter's water provides perhaps a quarter of the steam lift in a creamed cake; the rest comes from baking powder, baking soda, or trapped air. Swap to ghee or oil and you've removed that quarter, which is why ghee-based cakes often call for an extra quarter-teaspoon of baking powder per cup of fat. The full mechanics of how chemical leavening interacts with fat type are covered in baking powder's piece, but the rule is: lose water, add gas. The increase has limits — too much chemical leavener leaves a metallic aftertaste and an over-aerated crumb that collapses on cooling — so beyond about a 25% bump, the better fix is to add liquid back to the recipe rather than push more leavener.

The fourth lever is temperature. Butter's water buffers the fat's temperature in the pan; remove it and the fat heats faster and browns quicker. Ghee, lard, and clarified butter all hit smoke point sooner than whole butter, which is why higher smoke point — browns faster in pan is a feature for searing and a hazard for slow-melting confit. The pancake application (database score: 20 substitutions tracked) illustrates this neatly — pancakes ask the fat to grease the griddle without burning, and ghee or clarified butter outperforms whole butter for high-volume cooking precisely because the water that would have spattered is already gone. The same principle inverts in low-and-slow applications: a confit or a slow gentle braise wants the buffering effect of butter's water, and substituting ghee there will scorch the milk solids on the pan bottom long before the protein has rendered.

There's a fifth lever for the dietary swaps that aren't trying to imitate butter's structure at all but to remove its dairy. Applesauce at 1/2 cup per cup butter is the classic moisture-only replacement: it brings water (about 88%) and pectin (which mimics some of butter's emulsion-binding) but zero fat, so the recipe gets denser, slightly chewier, and less browned. The Maillard reaction needs both fat and protein to work fully, and applesauce has neither — so an applesauce-substituted brownie will be moist and pale, an applesauce-substituted cookie will be cakey rather than chewy. The 50% reduction in fat is the trade-off, and recipes designed around it (low-fat banana bread, healthier muffins) are designed knowing the texture and color will shift. For lactose-intolerant cooks, ghee remains the cleanest 1:1 option because the casein is removed in clarification; for fully dairy-free baking, oil-plus-water at the right ratio is closer to a true imitation than any plant butter, and many of the better commercial vegan butters are explicitly engineered to hit that 80/18 ratio with a mix of coconut oil, palm fraction, and emulsified water.

The takeaway across all four levers: a butter swap is rarely just a fat swap. It's a moisture-management decision that ripples through sugar, leavening, and temperature. Get the water right, and the rest of the recipe usually holds.

Related substitutions on SwapCook

For the full ranked list of butter alternatives by use case, see the butter substitute hub, and for application-specific guidance, the butter substitutes for baking and butter swaps in pie crust pages walk through which fats survive each recipe's water demands.